This study investigated whether two different writing tasks (narrative and argumentative), elicit different lexical features. This study also attempted to identify the relationship between lexical features and the level of writing proficiency. Seventyeight university students wrote one narrative and one argumentative essay, with a 30-minute time limit for each task. A total of 156 essays were rated by eight raters using holistic rubrics, and then were divided into three levels of writing proficiency. The written products were analyzed for nine lexical variables: mean word length,Type/Token Ratio, Root TTR (RTTR), Corrected TTR (CTTR), lexical density, the most frequent 1000-word families, the second most frequent 1000-word families,academic word list, and lexical frequency. The main findings are twofold. First,paired t-test analysis revealed significant differences between the two discourse modes in eight lexical variables. Second, the results of one-way ANOVA, Welch ANOVA, and Pearson’s correlation analysis indicated that RTTR and CTTR best discriminated between the different writing proficiency groups.